Dark corner in all my photos

[Jairo]

Hi,

I made 1 min exposures with of the Eagle Nebula, and all my photos had a right lower corder too dark.

I applied dark frames, which also have this, but it didn't cancel the problem.

There are no bias frames, but could it do this much? Do you have any idea of what is causing this?

Thanks,

Eagle Blue Deepskystack.jpgEagle G Deepskystack.jpgEagle Luminance Deepsky.jpgEagle Red Deepskystacker.jpgdarkframe 260 600006 23-11-55.jpg

[Amber Robot]

I couldn't really see it in your photos, but could it be a vignetting problem? What optical system are you using? If it's vignetting then a dark frame wouldn't show it. Does a flat-field frame show it?

[Jairo]

Hi,

The problem appears when I adjust the levels to bring out the details. I put the darker images because I thought you would preffer to see the originals.

I'm using a small 1/4" chip, with a SC8" f/10 with a f/3.3 focal reducer. So I don't think it's vignetting.

I haven't made a flat-field, mainly because I've never felt the need for it before. Previous 1 min expositions didn't show this.

[Hornblower]

I can't see anything amiss in any of those images. I would try a featureless expanse of daytime sky. If one corner looks distinctly darker than the others, I would suspect vignetting with faulty collimation.

How wide is the field of view in these images?

[Jairo]

I estimated 20x15 arcmin. I will try an artificial flatframe to

[Jairo]

There are more pronounced examples. After stacking.

Dumbbell R.jpgDumbbell G.jpgDumbbell B.jpg

[RickJ]

Use darks (to get rid of dark columns and other issues) and flats to get rid of the dark corner. You can't do deep sky imaging without these two steps or else both these issues will show up.

Rick

[Jairo]

I've read it's possible to make an artificial flat to cancel dark corners (but not dust marks, of course): I should get a light frame, aply a median noise filter, and blur the result. It worked marginally. I learned my lesson here.

I've used flats when stacking images of the Moon to cancel dust marks. I just unfocused the image. They worked so well I could barely see the dark blobs in individual light frames. So I got curious.

I'm told that a flat should be done so its intensity reach "around" 2/3 of the maximum of the histogram. Well, with so many hard constraints in deep space photography, why the intensity can be so poorly defined and still work fine? I thought we would need at least to fine tune a weighting factor for it.

[RickJ]

Flats should be raw, no processing of any kind other than bias subtraction. Once bias subtracted as many as possible (I'd use 20 or more with a video camera due to its high read noise) should be mean or median combined into a master flat. With a ABG chip such as in a video camera you want to keep the flat intensity in the linear part of the range. If you don't measure this, keep the intensity below 50%. Exact value is immaterial, just that it be low enough to be in the linear range of the chip. Fainter requires more flats be included to make the master however. I've used 5% and 50 of them quite successfully with a much lower noise camera. It isn't the brightness of the flat that's important but how well each pixel matches the true response of the camera in the optical train. So take flats without changing anything. Any rotation of the camera, change in focus position, alignment etc. will invalidate the flat. It is a precise process just that actual intensity is immaterial.

You need to understand how a flat works. It is a process in which relative intensity not actual intensity is the key. This is why it must be in the linear part of the response curve and why no adjustment to a pixel's intensity such as by blurring or smoothing is applied. The response of each pixel in the master is compared to the mean intensity of the master. So if the master has a mean value of say 25000 but a particular pixel, say 0,0 is 21,000 then the value of the images 0,0 pixel is divided by .84 (21,000/25,000) to obtain its corrected value. If the value of the pixel in the image is 1200 it becomes 1200/.84 = 1429. Now you should see why the optical train must exactly match when the image and flat are taken. If they don't the correction will be wrong and cause more problems than it solves.

For a crude correction of vignetting but not other irregularities cured by flats a program such as "GradientXTerminator" (PhotoShop plug-in) can do the job. No where near as well as a good flat and worthless for scientific measurement purposes it can save an image for which no flat is available.

The various articles at this link will help:
http://starizona.com/acb/ccd/ccd.aspx

Rick

[Jairo]

An additional information:

I noticed the dark corner is always there, no matter what configuration of filters, focal reducer and even camera rotation. It is always aligned with the sensor. Is that normal for a thing that should be corrected by a flat?

[RickJ]

Yes. This is their purpose. All optics have fall off. Can't be avoided. The edge of even a small frame sees the objective as slightly oval due to not being on axis. This means it gets less light. Flats correct this. They correct for other problems as well. Dust shadows, pixel sensitivity variation and more are also corrected. It is important each configuration have its own flat. Change any little thing, no matter how minor, and you'll need new flats since you can't normally mount the camera, filters etc. exactly the same each time. Dust can move every time the system is opened as well. Once you understand how they work as explained above you'll see they have to work if done correctly.

Rick

[Jairo]

It is important each configuration have its own flat. Change any little thing, no matter how minor, and you'll need new flats since you can't normally mount the camera, filters etc. exactly the same each time.Rick

I expected this too. My surprise is that it's always the same, no matter what optical configuraton I use: a dark corner on the lower right.

And about dark frames: the brightest pixels on my dark frames end up causing black pixels in the light frames. Is there a way to simple ignore that pixels, instead of making them count as black ones?

[RickJ]

If your camera has stuck pixels (those that are stuck at full on) then a dark will turn those black. These are quite rare in today's camera's however. Used to be quite common. In any case there are two solutions. Best is to use dithering when taking the image so these are in a different place each frame. Then use a stacking routine that rejects noise such as a type of Sigma reject combine. Min Max is good for this problem for instance. If you don't have enough frames for this type of combine to to work effectively (needs at least 6 with 9 or more best though CCDWare's CCDStack has a routine that works with 4 fairly well) then use a bad pixel map. Each processing software gives it a different name (CCDSoft calls it hot and cold pixel repair for instance) but it basically will detect pixels that stand out as either brighter or darker than those around it by a certain amount (most let you set the threshold). Then the software replaces the pixel value with an average of the value for the 8 pixels that surround it. Not as good as the first solution which replaces the pixel with the actual value that's an average of that pixels value on the good frames but does hide the problem.

A dark corner that is caused not by light coming into the camera but by that portion of the chip being of less sensitivity (rare) would also be cured by a flat since it is seen in the the flat as well. More likely such an apparent dark corner is really due to the rest of the frame being too light due to what is called amp glow. This is common with video cameras used for longer than a fraction of a second exposure but shouldn't be seen to this extent in a true deep sky camera. Though less expensive ones will have amp glow in one corner or edge (usually upper left or entire left side) rather than across the chip. In either case darks should remove this as it is a build up of dark current caused by warmth from the camera's electronics that is very consistent with exposure. Taking darks at the same time and temperature removes it nicely. It can be a bit more of a problem with cameras that don't regulate cooling and use chips with high dark current. Then a gradient removal tool will fix what darks (because they can't be perfectly matched to lights) don't remove. Usually only needed with cameras used beyond their design like a video camera for deep sky use.

Rick